University Institute of Technology, Burdwan University

A successful first Global Dialogue on AI Governance would be marked by its ability to translate diverse perspectives into a shared foundation for collective action. At present, approaches to AI governance remain fragmented, so achieving broad alignment on core principles such as safety, fairness, accountability, and transparency would represent an important step forward. However, success should extend beyond consensus-building to include the identification of practical, implementable measures, including policy recommendations or pilot frameworks that stakeholders are willing to adopt or test. Equally important is ensuring inclusivity in the dialogue, with meaningful participation from developing countries, students, researchers, and emerging innovators, alongside governments and industry leaders, so that governance frameworks reflect global realities rather than a limited set of interests. The dialogue should also emphasize a balanced approach that enables innovation while addressing potential risks in a responsible manner. Finally, establishing mechanisms for continuity, such as working groups, timelines, and follow-up processes, would be essential to ensure that the outcomes of the dialogue lead to sustained collaboration and measurable progress. Ultimately, its success would lie in its ability to move from discussion to action, fostering a shared sense of responsibility in shaping the future of AI.

• Safe, secure and trustworthy AI
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Interoperability of governance approaches

My selection of these four thematic areas is grounded in the research presented in my paper, which identifies the current fragmentation in AI governance as a systemic risk to global stability and equitable innovation. In this context, safe, secure and trustworthy AI, together with transparency, accountability and human oversight, constitute the technical foundation. Without standardized approaches to transparency and clearly defined lines of human responsibility, trust remains largely conceptual rather than operational, underscoring the need for verifiable safety standards, particularly to mitigate cross-border risks. Interoperability of governance approaches emerges as an urgent priority, as the proliferation of divergent national regulations may create governance gaps that can be exploited; my work therefore examines how interoperable frameworks can accommodate diverse national contexts while maintaining a shared global baseline for safety. At the same time, it is essential to recognize that technical considerations cannot be separated from their broader social context, and prioritizing the social, economic, ethical, cultural, linguistic and technical implications of AI ensures that governance frameworks remain inclusive, particularly for the Global South and underrepresented communities. Collectively, these priorities reflect a necessary shift from high-level principles to actionable and coordinated policy approaches, and my participation in this Dialogue is intended to contribute evidence-based insights toward the development of practical and implementable international standards.

While the listed themes cover the core pillars of current AI governance, I believe there are three emerging areas that need more explicit attention in the Dialogue. First is the compute and energy nexus, because AI development today is deeply tied to environmental impact, and the growing energy and water demands of training advanced models are creating a real sustainability challenge. Without some form of global alignment on green compute and resource efficiency, AI progress could start conflicting with broader climate goals. Second is the rise of autonomous agentic AI. We are no longer just dealing with systems that process information, but with systems that can take actions across digital and even physical environments. Current governance frameworks are still focused on static models, and there is a clear gap when it comes to defining accountability and liability in cases where human oversight is indirect or delayed. Third, data sovereignty and cultural commons is becoming increasingly important, especially for the Global South. There is a growing concern that data from these regions is being used to train models that neither represent nor benefit them, which risks creating a form of digital imbalance. So, there is a need to think of high-quality data not just as a resource to extract, but as something that carries cultural and social value and should be protected accordingly. Addressing these cross-cutting issues would help ensure that the Dialogue is not only responding to current challenges but is also preparing for what is coming next, making AI development more sustainable, inclusive, and genuinely human-centric.

The impact of gaps in AI governance is very real from my perspective as a student in India's academic and technical ecosystem. One of the biggest challenges is regulatory fragmentation. For a country like India, which is a major hub for IT and software development, the absence of a common international compliance baseline means that developers and researchers have to deal with a patchwork of global regulations. For student-led startups or academic projects, this becomes a serious barrier, because most of us do not have access to the legal support needed to ensure cross-border compliance. Alongside this, the accountability gap in autonomous systems creates additional uncertainty. As AI systems become more agent-like, there is still no clear global clarity on liability, and this makes it difficult for young researchers to move their work from the lab into real-world applications, especially in sensitive domains where the risks are not well defined. At the same time, I see a strong opportunity here. India is in a unique position to help bridge these gaps, especially by building on its experience with Digital Public Infrastructure and advocating similar open, interoperable approaches for AI governance. There is real potential for the academic community to contribute through frugal AI innovation and open-source safety tools that can work across borders. If leveraged well, this could contribute to building a more collaborative global AI ecosystem, where compute resources and diverse, high-quality datasets are shared more equitably. For someone like me, this creates an opportunity to be part of a governance-by-design approach, where academic work does not just stay theoretical but actively shapes more inclusive and transparent global standards.

The AI Dialogue can play a meaningful role in strengthening international cooperation if it moves beyond high-level discussions and connects directly with the realities of the IT sector and academic research. As a student in India's growing tech ecosystem, I see this Dialogue acting as a bridge in three important ways. First, it can help reduce the fragmentation in global regulations by encouraging alignment between different frameworks, which would make it easier for student-led startups and academic projects to build for a global audience without facing constant compliance challenges. Second, it can support more equal access to research opportunities by promoting the idea of a shared global AI ecosystem, where datasets, knowledge, and even compute resources are more accessible. This would allow students and researchers from developing countries to contribute to advanced AI work instead of being limited by resources. Third, it can bring much-needed clarity around accountability, especially as AI systems become more autonomous. Having clearer global understanding of responsibility and liability would make it easier for young developers to move their work from research to real-world applications with confidence. Overall, if the Dialogue focuses on these practical aspects, it can move from being just a platform for discussion to something that actively shapes a more inclusive, collaborative, and workable global approach to AI governance.

Building on existing initiatives is essential if the Dialogue is to create real impact rather than start from scratch. In that context, efforts like the Global Partnership on AI can be valuable, especially given India's active role, by using its technical working groups on areas like data governance and AI safety to better connect research with actual policy outcomes. Similarly, the concept of Digital Public Infrastructure offers a strong foundation, particularly through models like India Stack, by showing that governance can be enabling rather than restrictive, creating open and inclusive systems that support innovation across diverse populations. The UNESCO Recommendation on the Ethics of AI already provides a widely accepted ethical baseline, and tools like its Readiness Assessment Methodology can help countries practically assess and improve their governance capacities. At the same time, platforms such as the OECD AI Policy Observatory remain important for tracking global trends and supporting evidence-based academic research. What makes this Dialogue distinct, however, is its universal legitimacy as a UN-led platform that brings together all member states, unlike more limited multilateral groups. For students and the broader IT and academic sectors, this creates real value by helping bridge the global gap, ensuring that governance frameworks reflect not just a few regions but also the realities of countries like India, including linguistic and economic diversity. It also has the potential to move beyond largely voluntary guidelines toward more coordinated and predictable global approaches, which would give much-needed clarity and confidence to young developers and researchers working on AI systems.

To ensure the AI Dialogue is genuinely inclusive and impactful, its format should move beyond traditional top-down diplomacy and adopt a more layered and participatory structure. One important step would be to include dedicated technical peer-review tracks, where academia and the broader technical community can submit research papers and benchmarks, ensuring that policy discussions are grounded in evidence rather than only consensus. At the same time, the Dialogue should adopt a regional hybrid model, where localized hubs in major tech centers such as those in India can contribute in real time to central discussions, making participation more accessible for students and stakeholders who may face travel constraints. There should also be a structured and standardized "call for inputs" specifically for youth and students, allowing them to share perspectives on long-term challenges like safety, employment, and societal impact, possibly through formats like governance sandboxes that showcase real-world use cases and their regulatory gaps. In addition, the Dialogue should actively promote public, private, and academic collaboration through dedicated task forces that focus on practical implementation, especially around how global guidelines can be integrated into existing IT systems and research environments. Overall, by adopting a more modular, decentralized, and collaborative approach, the Dialogue can evolve from a series of discussions into a continuous ecosystem that meaningfully connects technical expertise with global policy-making.

One of the most important gaps in global AI governance discussions today is that many of the people most affected by AI are still not meaningfully represented in decision-making spaces. Voices from the Global South, including countries like India, are often underrepresented when it comes to shaping technical standards and policy frameworks, and within these regions, students, early-stage researchers, and small startup founders are rarely included despite actively building AI in diverse and resource-constrained environments. At the same time, informal workers and communities impacted by automation, as well as linguistic and cultural minorities, are often overlooked, which results in systems that do not fully reflect or serve them. Addressing this requires moving beyond symbolic inclusion to structural participation, such as creating dedicated tracks for youth and grassroots innovators, enabling hybrid and funded access to reduce participation barriers, and establishing regional hubs through universities and local institutions. It is also important to support multilingual engagement so that contributions are not limited by language, and to introduce mechanisms like open calls or governance sandboxes where underrepresented groups can share real-world use cases and challenges. By embedding these approaches into the design of the Dialogue, AI governance can become more representative, grounded, and aligned with the realities of a truly global population.

To move beyond static speeches and create a more engaging and outcome-driven environment, the AI Dialogue should adopt interactive and practice-oriented formats. One approach could be AI governance "red-teaming" simulations, where participants, including students and technical experts, actively test proposed governance frameworks against realistic AI-driven scenarios, helping identify gaps between policy intent and real-world implementation. In addition, global policy hackathons could bring together cross-disciplinary teams of developers, legal experts, and researchers to build practical tools for areas like compliance, transparency, and cross-border data sharing, turning discussions into tangible outputs. Introducing reverse mentorship sessions would also add value by pairing experienced policymakers with students and early-stage researchers, especially from regions like the Global South, allowing those working directly with AI systems to share ground-level insights on emerging challenges. Another impactful format would be live policy sandboxes, where real-world case studies, such as AI applications in public digital systems, are explored collaboratively, enabling stakeholders to experiment with how such models could be governed or scaled within an international framework. By incorporating these kinds of participatory and simulation-based formats, the Dialogue can move from passive exchange to active problem-solving, generating practical insights while also fostering a stronger sense of shared ownership among both current and future stakeholders in AI governance.

To address the challenges of fragmentation and inequity in AI governance, the Dialogue should build on models that have already shown practical success. One strong example is India's Digital Public Infrastructure approach, including systems like Aadhaar and UPI, which demonstrate how foundational digital systems can be designed as public goods rather than closed, proprietary platforms. Extending this thinking to AI by treating elements such as compute access and high-quality datasets as shared infrastructure could make AI development more inclusive while also improving transparency and oversight. Another important model is the use of regulatory sandboxes, which allow students, researchers, and startups to test AI applications in controlled environments under regulatory supervision, helping balance innovation with risk management before large-scale deployment. The OECD AI Policy Observatory also provides a valuable example of evidence-based governance by offering a continuously updated view of national AI strategies and policy trends, enabling countries to learn from each other and move toward greater alignment. In addition, multi-stakeholder red-teaming initiatives, such as those led by national AI safety institutes, show how collaboration between technical experts and policymakers can proactively identify risks in advanced systems rather than responding after harm occurs. By building on and connecting these approaches, the AI Dialogue has the opportunity to move beyond broad principles and help shape a more practical, interoperable, and globally inclusive governance framework.